Resident EducationThe Effect of Teaching Search Strategies on Perceptual Performance
Introduction
Perceptual errors account for a substantial part of misdiagnoses in radiology (1) and can be related to the search behavior of the observer (2). For educational purposes, it is important to identify which visual search patterns are most effective and to investigate if teaching search strategies improves perception.
Visual search characteristics that are related to expertise and high performance have been identified in various radiology perception tasks (3). For example experts tend to fixate on abnormalities faster 4, 5, 6 and need less time and a smaller number of eye fixations to inspect the image 7, 8. These characteristics derive from experience, and they lack an underlying structure that can be taught to novices.
Some specific visual search patterns are found to be related to high performance 4, 9, 10, 11. Most patterns apply to visual search in X-rays, such as chest X-rays or mammography. Two visual search types are distinguished for searching chest computed tomography (CT) images: “scanners” and “drillers” (11). Scanners tend to visually search a single slice, before scrolling further through the stack, whereas drillers focus their eyes on one quadrant of the lung fields and quickly scroll through the stack in depth before moving to another quadrant. Drillers outperformed scanners with respect to higher true positive rates and a larger lung coverage (11). One interesting finding was that, when given the option to search freely, more experienced readers tend to select “drilling” as a search pattern (the more effective pattern), suggesting it might be a pattern that has, consciously or unconsciously, evolved through instruction or practice. The relationship between search patterns and experience has been noted in several other studies 4, 9, 10, 11, although it is unknown if experts unconsciously adopt these patterns or deliberately chose or had acquired one, as a strategy.
Teaching junior trainees to use expert search strategies may not necessarily be effective. First, learning the strategy may not be easy, particularly given that most experts acquire their behaviors after years of practice. Second, the improvement in perceptual performance that comes with experience is probably due to multiple factors. Knowledge gained and feedback received are known to be critical factors in developing visual expertise 12, 13, 14, 15. Therefore, it is not evident that learners' perceptual performance will improve simply by using the search strategies of experts. However, some perceptual tasks, such as finding lung nodules on chest CT scans, do not depend on a large knowledge base, and therefore teaching a search strategy may improve detection. Experimental studies may be beneficial to determine if search patterns can be taught to junior observers, and if this can improve perceptual performance.
The aim of this research study is twofold: (1) to investigate if drilling and scanning search strategies can be taught to junior radiology trainees, and (2) to compare the effect of teaching each search strategy on trainee's perception accuracy of lung nodule detection. We hypothesized that junior radiology trainees could adopt a new search strategy after instruction and that the use of a drilling strategy would improve the trainees' perceptual performance compared to a scanning strategy.
Section snippets
Design
An experimental study was conducted to compare the effect of two teaching methods on perceptual performance. A randomized crossover design was chosen to adjust for individual variation in performance, differences in search strategies prior to any search strategy instruction, and possible differences in search behavior due to the sequence of the search strategy instructions. The design is illustrated in Figure 1.
Study Population and Procedure
Over a 3-month period, 19 (70%) first and second-year radiology trainees of a US
Participants
After randomization, 9 participants were assigned to group A and 10 to group B, of which 4 in each group were second-year residents.
Test Performance
Reliabilities of the tests were acceptable: Cronbach's α were 0.74, 0.82, and 0.84 for Tests 1, 2, and 3, respectively. Average P values were 0.49, 0.39 and 0.48, respectively.
Scroll Behavior
Median number of runs and scrolling time for each search strategy are provided in Table 2.
There was no significant effect of study group or year of residency on the number of runs or time.
Discussion
Search strategy instruction had a significant effect on both scroll behavior and perceptual performance. The scanning instruction decreased the number of long scroll movements and scrolling time. The drilling instruction did not alter scroll behavior significantly. However, the majority of participants reported already using some kind of drilling strategy at their free search. Perceptual performance following drilling search instructions outperformed performance following scanning search
Conclusion
Search strategy instruction can influence scroll behavior and perceptual performance of junior radiology residents completing a lung nodule detection task. In junior trainees, a drilling strategy yields a better perceptual performance than a scanning strategy. Teaching a scanning strategy further decreases the perceptual performance of junior radiology residents below their baseline performance. Teaching a drilling strategy for detecting lung nodules in chest CT scans is therefore preferable.
References (18)
Visual scanning patterns of radiologists searching mammograms
Acad Radiol
(1996)- et al.
How do radiologists do it? The influence of experience and training on searching for chest nodules
Radiography
(2006) - et al.
Searching for bone fractures: a comparison with pulmonary nodule search
Acad Radiol
(1994) - et al.
How experience and training influence mammography expertise
Acad Radiol
(1999) - et al.
Nature of expertise in searching mammograms for breast masses
Acad Radiol
(1996) - et al.
Common patterns in 558 diagnostic radiology errors
J Med Imaging Radiat Oncol
(2012) - et al.
Visual scanning, pattern recognition and decision-making in pulmonary nodule detection
Invest Radiol
(1978) - et al.
How visual search relates to visual diagnostic performance: a narrative systematic review of eye-tracking research in radiology
Adv Health Sci Educ Theory Pract
(2016) - et al.
The assessment of stroke multidimensional CT and MR imaging using eye movement analysis: does modality preference enhance observer performance?
Cited by (7)
How to Read an Abdominal CT: Insights from the Visual and Cognitive Sciences Translated for Clinical Practice
2022, Current Problems in Diagnostic RadiologyCitation Excerpt :Even with the advent of AI, you will still need to be able to search with your own prepared and informed intelligence. Acquiring perceptual (visual) fluency typically requires years of training and experience, but there is evidence that the acquisition of this skill can be accelerated through nonverbal, implicit learning processes, such as image immersion.41,55 Acquisition can be accelerated, for instance, through trainings on 60 images viewed 3 times each day for 4 days, and 1280 images for 52 minutes.7
Towards expert gaze modeling and recognition of a user's attention in realtime
2020, Procedia Computer ScienceA Review of Perceptual Expertise in Radiology-How it develops, How we can test it, and Why humans still matter in the era of Artificial Intelligence
2020, Academic RadiologyCitation Excerpt :Eckstein et al. found higher detectability for calcifications in single slice images and relatively improved detection of masses in volumetric imaging (39). Using saliency maps, Wen et al. demonstrated that the great majority of radiologists (42) use dynamic (motion based) information when interpreting cross-sectional imaging (43). Both Wen and Eckstein studies lend support to the notion that observer performance in 2D search might not generalize to volumetric search tasks (39,43), as the 2 types of search entail different perceptual tasks.
Improving Novice Radiology Trainees’ Perception Using Fine Art
2017, Journal of the American College of RadiologyCitation Excerpt :To illustrate this, recent studies have shown that radiology-specific perception training can improve the detection of lung nodules. Van der Gijp et al [9] taught two groups of residents different perception strategies for detecting lung nodules. They found that residents were more likely to correctly identify lung nodules in a more efficient manner with one of the perception methods.
Characteristics of expert search behavior in volumetric medical image interpretation
2021, Journal of Medical ImagingThe search patterns of abdominal imaging subspecialists for abdominal computed tomography: Toward a foundational pattern for new radiology residents
2021, Journal of Clinical Imaging Science